derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
78 lines
2.6 KiB
LLVM
78 lines
2.6 KiB
LLVM
; RUN: llc -ppc-asm-full-reg-names -verify-machineinstrs < %s \
|
|
; RUN: -mtriple=powerpc64le-unknown-linux-gnu | FileCheck %s -check-prefix=INST
|
|
; RUN: llc -ppc-asm-full-reg-names -verify-machineinstrs -ppc-lsr-no-insns-cost=true \
|
|
; RUN: < %s -mtriple=powerpc64le-unknown-linux-gnu | FileCheck %s -check-prefix=REG
|
|
|
|
; void test(unsigned *a, unsigned *b, unsigned *c)
|
|
; {
|
|
; for (unsigned long i = 0; i < 1024; i++)
|
|
; c[i] = a[i] + b[i];
|
|
; }
|
|
;
|
|
; compile with -fno-unroll-loops
|
|
|
|
define void @lsr-insts-cost(ptr %0, ptr %1, ptr %2) {
|
|
; INST-LABEL: lsr-insts-cost
|
|
; INST: .LBB0_4: # =>This Inner Loop Header: Depth=1
|
|
; INST-NEXT: lxvd2x vs34, r3, r6
|
|
; INST-NEXT: lxvd2x vs35, r4, r6
|
|
; INST-NEXT: vadduwm v2, v3, v2
|
|
; INST-NEXT: stxvd2x vs34, r5, r6
|
|
; INST-NEXT: addi r6, r6, 16
|
|
; INST-NEXT: bdnz .LBB0_4
|
|
;
|
|
; REG-LABEL: lsr-insts-cost
|
|
; REG: .LBB0_4: # =>This Inner Loop Header: Depth=1
|
|
; REG-NEXT: lxvd2x vs34, 0, r3
|
|
; REG-NEXT: lxvd2x vs35, 0, r4
|
|
; REG-NEXT: addi r4, r4, 16
|
|
; REG-NEXT: addi r3, r3, 16
|
|
; REG-NEXT: vadduwm v2, v3, v2
|
|
; REG-NEXT: stxvd2x vs34, 0, r5
|
|
; REG-NEXT: addi r5, r5, 16
|
|
; REG-NEXT: bdnz .LBB0_4
|
|
%4 = getelementptr i32, ptr %2, i64 1024
|
|
%5 = getelementptr i32, ptr %0, i64 1024
|
|
%6 = getelementptr i32, ptr %1, i64 1024
|
|
%7 = icmp ugt ptr %5, %2
|
|
%8 = icmp ugt ptr %4, %0
|
|
%9 = and i1 %7, %8
|
|
%10 = icmp ugt ptr %6, %2
|
|
%11 = icmp ugt ptr %4, %1
|
|
%12 = and i1 %10, %11
|
|
%13 = or i1 %9, %12
|
|
br i1 %13, label %28, label %14
|
|
|
|
14: ; preds = %3, %14
|
|
%15 = phi i64 [ %25, %14 ], [ 0, %3 ]
|
|
%16 = getelementptr inbounds i32, ptr %0, i64 %15
|
|
%17 = bitcast ptr %16 to ptr
|
|
%18 = load <4 x i32>, ptr %17, align 4
|
|
%19 = getelementptr inbounds i32, ptr %1, i64 %15
|
|
%20 = bitcast ptr %19 to ptr
|
|
%21 = load <4 x i32>, ptr %20, align 4
|
|
%22 = add <4 x i32> %21, %18
|
|
%23 = getelementptr inbounds i32, ptr %2, i64 %15
|
|
%24 = bitcast ptr %23 to ptr
|
|
store <4 x i32> %22, ptr %24, align 4
|
|
%25 = add i64 %15, 4
|
|
%26 = icmp eq i64 %25, 1024
|
|
br i1 %26, label %27, label %14
|
|
|
|
27: ; preds = %14, %28
|
|
ret void
|
|
|
|
28: ; preds = %3, %28
|
|
%29 = phi i64 [ %36, %28 ], [ 0, %3 ]
|
|
%30 = getelementptr inbounds i32, ptr %0, i64 %29
|
|
%31 = load i32, ptr %30, align 4
|
|
%32 = getelementptr inbounds i32, ptr %1, i64 %29
|
|
%33 = load i32, ptr %32, align 4
|
|
%34 = add i32 %33, %31
|
|
%35 = getelementptr inbounds i32, ptr %2, i64 %29
|
|
store i32 %34, ptr %35, align 4
|
|
%36 = add nuw nsw i64 %29, 1
|
|
%37 = icmp eq i64 %36, 1024
|
|
br i1 %37, label %27, label %28
|
|
}
|